Hi,
I just treated myself to a Geforce FX 5600, and was dismayed to find that Croquet segfaults when trying to render t-meshes. The segfault doesn't happen immediately; many frames may be rendered successfully before it crashes. No Squeak3D.log is left behind after such a crash.
I'm using NVIDIA's 4363 drivers (haven't updated to the newest ones yet; they stopped distributing in .tar.gz form) on a 2.5.75 kernel. The same driver worked fine with Croquet and my previous card, a Geforce 2 MX200.
Squeak it the only app that I can find that exhibits a problem. I can run the OpenGL screensavers, all of the demo from NVIDIA's SDK, Tux Racer, and Quake2 just fine.
The following is what is spit out on my command line. I'll try to look into it further, but thought I might save myself some time if any other Linux Croqueteers have encountered this.
Thanks, Joshua
Segmentation fault
1105393924 GLX>drawIndexed:faces:vertices:normals:colors:texCoords: 1105393832 GLX>drawIndexedTriangles:vertices:normals:colors:texCoords: 1105392508 TMesh>renderPrimitive:alpha: 1105391824 TMesh>renderPrimitive: 1105391732 TMesh>render: 1105391640 TFrame>renderFrame: 1105391036 [] in TFrame>renderFrame: 1105388572 OrderedCollection>do: 1105388436 TFrame>renderFrame: 1105387932 TSpace>renderSpace:depth: 1105381668 TSpace>renderSpace 1105381576 TeapotMorph>glRenderOn: 1105381484 TeapotMorph>renderOn: 1105381392 TeapotMorph>drawAcceleratedOn: 1105379644 TeapotMorph>drawOn: 1105379552 Canvas>draw: 1105379460 Canvas>drawMorph: 1105378724 [] in Morph>fullDrawOn:
Is this with the default models? Any idea what was on the screen at the time? Croquet is pretty good about culling, so if you don't see it, it probably isn't rendering.
DAS
At 01:03 AM 8/2/2003 -0400, you wrote:
Hi,
I just treated myself to a Geforce FX 5600, and was dismayed to find that Croquet segfaults when trying to render t-meshes. The segfault doesn't happen immediately; many frames may be rendered successfully before it crashes. No Squeak3D.log is left behind after such a crash.
I'm using NVIDIA's 4363 drivers (haven't updated to the newest ones yet; they stopped distributing in .tar.gz form) on a 2.5.75 kernel. The same driver worked fine with Croquet and my previous card, a Geforce 2 MX200.
Squeak it the only app that I can find that exhibits a problem. I can run the OpenGL screensavers, all of the demo from NVIDIA's SDK, Tux Racer, and Quake2 just fine.
The following is what is spit out on my command line. I'll try to look into it further, but thought I might save myself some time if any other Linux Croqueteers have encountered this.
Thanks, Joshua
Segmentation fault
1105393924 GLX>drawIndexed:faces:vertices:normals:colors:texCoords: 1105393832 GLX>drawIndexedTriangles:vertices:normals:colors:texCoords: 1105392508 TMesh>renderPrimitive:alpha: 1105391824 TMesh>renderPrimitive: 1105391732 TMesh>render: 1105391640 TFrame>renderFrame: 1105391036 [] in TFrame>renderFrame: 1105388572 OrderedCollection>do: 1105388436 TFrame>renderFrame: 1105387932 TSpace>renderSpace:depth: 1105381668 TSpace>renderSpace 1105381576 TeapotMorph>glRenderOn: 1105381484 TeapotMorph>renderOn: 1105381392 TeapotMorph>drawAcceleratedOn: 1105379644 TeapotMorph>drawOn: 1105379552 Canvas>draw: 1105379460 Canvas>drawMorph: 1105378724 [] in Morph>fullDrawOn:
This happened even in a TeaTestMorph in a 364 image; the only things there are a mirror, the floor, and a particle system. It crashes drawing a frame.
However, it could have been a driver bug, because I installed the 4496 drivers and it works great now. If you're still interested in tracking it down, I can reinstall the old drivers, but my personal itch has been scratched.
Thanks, Joshua
BTW, anyone who wants to get the .tar.gz out of the .rpm files without installing the package, use:
rpm2cpio NVIDIA-whatever.rpm | cpio -i
If you want to run the drivers with a 2.5/2.6 linux kernel, a patch is available at:
http://www.minion.de/nvidia.html
On Sat, Aug 02, 2003 at 01:32:28AM -0400, David A. Smith wrote:
Is this with the default models? Any idea what was on the screen at the time? Croquet is pretty good about culling, so if you don't see it, it probably isn't rendering.
DAS
At 01:03 AM 8/2/2003 -0400, you wrote:
Hi,
I just treated myself to a Geforce FX 5600, and was dismayed to find that Croquet segfaults when trying to render t-meshes. The segfault doesn't happen immediately; many frames may be rendered successfully before it crashes. No Squeak3D.log is left behind after such a crash.
I'm using NVIDIA's 4363 drivers (haven't updated to the newest ones yet; they stopped distributing in .tar.gz form) on a 2.5.75 kernel. The same driver worked fine with Croquet and my previous card, a Geforce 2 MX200.
Squeak it the only app that I can find that exhibits a problem. I can run the OpenGL screensavers, all of the demo from NVIDIA's SDK, Tux Racer, and Quake2 just fine.
The following is what is spit out on my command line. I'll try to look into it further, but thought I might save myself some time if any other Linux Croqueteers have encountered this.
Thanks, Joshua
Segmentation fault
1105393924 GLX>drawIndexed:faces:vertices:normals:colors:texCoords: 1105393832 GLX>drawIndexedTriangles:vertices:normals:colors:texCoords: 1105392508 TMesh>renderPrimitive:alpha: 1105391824 TMesh>renderPrimitive: 1105391732 TMesh>render: 1105391640 TFrame>renderFrame: 1105391036 [] in TFrame>renderFrame: 1105388572 OrderedCollection>do: 1105388436 TFrame>renderFrame: 1105387932 TSpace>renderSpace:depth: 1105381668 TSpace>renderSpace 1105381576 TeapotMorph>glRenderOn: 1105381484 TeapotMorph>renderOn: 1105381392 TeapotMorph>drawAcceleratedOn: 1105379644 TeapotMorph>drawOn: 1105379552 Canvas>draw: 1105379460 Canvas>drawMorph: 1105378724 [] in Morph>fullDrawOn:
Sounds like the same problem others have been experiencing, even with older boards. Another driver helped, sometimes. It would be great if this new driver now indeed solves this issue for good.
Maybe you could add a little blurp to
http://minnow.cc.gatech.edu/squeak/CroquetLinux
And could you verify the info there is up-to-date?
-- Bert
Joshua 'Schwa' Gargus wrote:
This happened even in a TeaTestMorph in a 364 image; the only things there are a mirror, the floor, and a particle system. It crashes drawing a frame.
However, it could have been a driver bug, because I installed the 4496 drivers and it works great now. If you're still interested in tracking it down, I can reinstall the old drivers, but my personal itch has been scratched.
Thanks, Joshua
BTW, anyone who wants to get the .tar.gz out of the .rpm files without installing the package, use:
rpm2cpio NVIDIA-whatever.rpm | cpio -i
If you want to run the drivers with a 2.5/2.6 linux kernel, a patch is available at:
http://www.minion.de/nvidia.html
On Sat, Aug 02, 2003 at 01:32:28AM -0400, David A. Smith wrote:
Is this with the default models? Any idea what was on the screen at the time? Croquet is pretty good about culling, so if you don't see it, it probably isn't rendering.
DAS
At 01:03 AM 8/2/2003 -0400, you wrote:
Hi,
I just treated myself to a Geforce FX 5600, and was dismayed to find that Croquet segfaults when trying to render t-meshes. The segfault doesn't happen immediately; many frames may be rendered successfully before it crashes. No Squeak3D.log is left behind after such a crash.
I'm using NVIDIA's 4363 drivers (haven't updated to the newest ones yet; they stopped distributing in .tar.gz form) on a 2.5.75 kernel. The same driver worked fine with Croquet and my previous card, a Geforce 2 MX200.
Squeak it the only app that I can find that exhibits a problem. I can run the OpenGL screensavers, all of the demo from NVIDIA's SDK, Tux Racer, and Quake2 just fine.
The following is what is spit out on my command line. I'll try to look into it further, but thought I might save myself some time if any other Linux Croqueteers have encountered this.
Thanks, Joshua
Segmentation fault
1105393924 GLX>drawIndexed:faces:vertices:normals:colors:texCoords: 1105393832 GLX>drawIndexedTriangles:vertices:normals:colors:texCoords: 1105392508 TMesh>renderPrimitive:alpha: 1105391824 TMesh>renderPrimitive: 1105391732 TMesh>render: 1105391640 TFrame>renderFrame: 1105391036 [] in TFrame>renderFrame: 1105388572 OrderedCollection>do: 1105388436 TFrame>renderFrame: 1105387932 TSpace>renderSpace:depth: 1105381668 TSpace>renderSpace 1105381576 TeapotMorph>glRenderOn: 1105381484 TeapotMorph>renderOn: 1105381392 TeapotMorph>drawAcceleratedOn: 1105379644 TeapotMorph>drawOn: 1105379552 Canvas>draw: 1105379460 Canvas>drawMorph: 1105378724 [] in Morph>fullDrawOn:
Thanks Bert, I updated the swiki page.
Joshua
On Sat, Aug 02, 2003 at 11:25:54AM +0200, Bert Freudenberg wrote:
Sounds like the same problem others have been experiencing, even with older boards. Another driver helped, sometimes. It would be great if this new driver now indeed solves this issue for good.
Maybe you could add a little blurp to
http://minnow.cc.gatech.edu/squeak/CroquetLinux
And could you verify the info there is up-to-date?
-- Bert
squeak-dev@lists.squeakfoundation.org